Using Bayesian Networks to Visually Compare the Countries: an Example from PISA
Seyfullah Tingir, MAE; Russell Almond, PhD
Abstract
Bayesian networks provide a graphical representation of a probability distribution where separation in the graph represents conditional independence of the corresponding variables. Finding a minimal Bayesian network, which expresses the conditional independence relationships among a set of variables, allows an analyst to visualize the relationship among the variables. This study applies a Bayesian network learning algorithm to data from the Program for International Student Assessment (PISA) 2012 mathematics literacy data. Particularly, the relationship among the self-concept, self-efficacy, math interest, and math achievement are examined for both the United States and Turkey. Comparing the Bayesian networks illustrates the different relationships between self-efficacy and math achievement for the students from the two countries.
Full Text: PDF